resnet101-base_tobacco-cnn_tobacco3482_kd_CEKD_t1.0_a0.5

This model is a fine-tuned version of bdpc/resnet101-base_tobacco on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.8577
• Accuracy: 0.53
• Brier Loss: 0.6406
• Nll: 2.1208
• F1 Micro: 0.53
• F1 Macro: 0.4957
• Ece: 0.3004
• Aurc: 0.3168

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.0001
• train_batch_size: 256
• eval_batch_size: 256
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Brier Loss	Nll	F1 Micro	F1 Macro	Ece	Aurc
No log	1.0	4	1.4267	0.05	0.9008	9.6592	0.0500	0.0177	0.1432	0.9439
No log	2.0	8	1.4006	0.155	0.8969	7.9140	0.155	0.0268	0.2365	0.9603
No log	3.0	12	1.4621	0.155	0.9457	13.3695	0.155	0.0268	0.3013	0.9107
No log	4.0	16	2.1836	0.155	1.3252	12.8977	0.155	0.0268	0.6400	0.7514
No log	5.0	20	2.4365	0.155	1.3998	8.4435	0.155	0.0268	0.7030	0.6102
No log	6.0	24	2.1554	0.155	1.2534	6.9190	0.155	0.0279	0.5987	0.6271
No log	7.0	28	1.5617	0.175	0.9637	5.7454	0.175	0.0462	0.3802	0.6485
No log	8.0	32	1.3267	0.245	0.8707	5.2368	0.245	0.0835	0.2961	0.5438
No log	9.0	36	1.2434	0.19	0.8886	5.0360	0.19	0.0471	0.3198	0.7720
No log	10.0	40	1.0721	0.305	0.8123	4.5157	0.305	0.1762	0.2684	0.5269
No log	11.0	44	1.1256	0.22	0.8429	3.9215	0.22	0.1083	0.2812	0.7346
No log	12.0	48	0.9865	0.35	0.7676	3.4553	0.35	0.2565	0.2884	0.4790
No log	13.0	52	1.0206	0.355	0.7899	3.3582	0.3550	0.2278	0.2954	0.5883
No log	14.0	56	0.9096	0.415	0.6994	3.2174	0.415	0.3147	0.2563	0.3596
No log	15.0	60	0.9187	0.415	0.7129	3.2059	0.415	0.2742	0.2941	0.3971
No log	16.0	64	0.8905	0.395	0.6956	2.9931	0.395	0.2618	0.2590	0.3826
No log	17.0	68	0.9108	0.425	0.7073	3.1634	0.425	0.2855	0.2995	0.3685
No log	18.0	72	0.8769	0.465	0.6706	3.1088	0.465	0.3652	0.2855	0.3261
No log	19.0	76	0.8585	0.475	0.6687	2.8710	0.4750	0.3884	0.2916	0.3282
No log	20.0	80	0.9822	0.405	0.7378	2.8889	0.405	0.3570	0.2850	0.4895
No log	21.0	84	0.9324	0.445	0.6992	2.7975	0.445	0.3553	0.3021	0.3762
No log	22.0	88	1.0330	0.42	0.7350	2.7487	0.4200	0.3506	0.2984	0.4771
No log	23.0	92	0.8755	0.455	0.6674	2.5903	0.455	0.3415	0.2570	0.3352
No log	24.0	96	0.8651	0.47	0.6443	2.8456	0.47	0.3800	0.2451	0.2975
No log	25.0	100	0.9567	0.445	0.7150	2.7083	0.445	0.3727	0.2667	0.4676
No log	26.0	104	1.0224	0.42	0.7376	2.4408	0.4200	0.3367	0.2968	0.5019
No log	27.0	108	0.8365	0.525	0.6407	2.6426	0.525	0.4496	0.2960	0.2657
No log	28.0	112	0.9798	0.425	0.7287	2.6379	0.425	0.3489	0.2640	0.4668
No log	29.0	116	0.9226	0.44	0.6965	2.5748	0.44	0.3669	0.2561	0.4054
No log	30.0	120	0.8303	0.49	0.6398	2.4839	0.49	0.3924	0.2981	0.2936
No log	31.0	124	0.8426	0.52	0.6478	2.5282	0.52	0.4322	0.3109	0.3084
No log	32.0	128	0.9111	0.45	0.6970	2.3870	0.45	0.3947	0.2837	0.4448
No log	33.0	132	0.8723	0.51	0.6524	2.6124	0.51	0.4170	0.2536	0.3365
No log	34.0	136	0.8936	0.47	0.6671	2.8892	0.47	0.3814	0.2436	0.3357
No log	35.0	140	1.2870	0.42	0.7660	4.4020	0.4200	0.3468	0.2860	0.4606
No log	36.0	144	0.9991	0.455	0.7289	2.6973	0.455	0.4132	0.3272	0.4684
No log	37.0	148	1.6352	0.365	0.8356	4.7695	0.3650	0.3020	0.3312	0.6069
No log	38.0	152	1.3014	0.39	0.8213	2.9436	0.39	0.3382	0.3262	0.5476
No log	39.0	156	1.0294	0.415	0.7361	2.7188	0.415	0.3446	0.2454	0.4632
No log	40.0	160	0.8825	0.52	0.6538	2.3887	0.52	0.4608	0.2721	0.3186
No log	41.0	164	0.8572	0.54	0.6288	2.4201	0.54	0.4822	0.2963	0.2899
No log	42.0	168	0.8393	0.535	0.6291	2.3587	0.535	0.4726	0.2824	0.2937
No log	43.0	172	0.8369	0.515	0.6303	2.4060	0.515	0.4583	0.2689	0.2903
No log	44.0	176	0.8458	0.49	0.6346	2.3323	0.49	0.4428	0.2526	0.2951
No log	45.0	180	0.8446	0.49	0.6367	2.2207	0.49	0.4289	0.2655	0.3041
No log	46.0	184	0.8324	0.54	0.6289	2.3685	0.54	0.4779	0.2571	0.2873
No log	47.0	188	0.8658	0.515	0.6486	2.3922	0.515	0.4584	0.2623	0.3100
No log	48.0	192	0.8516	0.525	0.6410	2.4448	0.525	0.4700	0.3006	0.3044
No log	49.0	196	0.8520	0.55	0.6350	2.2049	0.55	0.4947	0.3030	0.2980
No log	50.0	200	0.8577	0.53	0.6406	2.1208	0.53	0.4957	0.3004	0.3168

Framework versions

• Transformers 4.36.0.dev0
• Pytorch 2.2.0.dev20231112+cu118
• Datasets 2.14.5
• Tokenizers 0.14.1